Dsacpo

sa

a

b

ce

4

Effect of Referring PhysicianSpecialty and Practice Type on

Referral for Image-Guided BreastBiopsy

Elizabeth Lazarus, MDa, Martha B. Mainiero, MDa, Ilana F. Gareen, PhDb

Purpose: To determine whether referring physician specialty and practice type affect the likelihood of referralfor image-guided breast biopsy.

Method: We reviewed the records of every patient who received a Breast Imaging Reporting and Data System(BI-RADS) category 4 or 5 examination (mammography, ultrasound, or both) performed during the year 2000at both our community- and hospital-based practices. We recorded the referring physician specialty and officelocation (academic medical center vs. community), breast-imaging facility location (hospital vs. community),patient age, patient insurance status, BI-RADS category, and palpability of the lesion. Multiple logisticregression analysis was performed to examine the relationship of these factors to the biopsy type (surgical vs.image-guided).

Results: Of the 831 patients with a BI-RADS Category 4 or 5 report, 734 underwent follow-up surgical orimage-guided needle biopsy. Multiple logistic regression analysis demonstrated that referral by a nonsurgeon orby a physician at the academic medical center was associated with a higher likelihood of the patient’s beingreferred for image-guided biopsy. Patients referred by surgeons and community physicians were more likely toundergo surgical biopsy. BI-RADS Category 5 examinations and palpable lesions were also associated with ahigher probability of undergoing surgical biopsy. Patient age, insurance status, and breast-imaging facilitylocation were not related to biopsy type.

Conclusions: In our practice, patients referred for breast imaging by nonsurgeons and academic physicianswere more likely to be referred for image-guided biopsy, whereas patients referred by surgeons and communityphysicians were more likely to undergo surgical biopsy.

Key Words: Breast, biopsy, health policy and practice

J Am Coll Radiol 2005;2:488-493. Copyright © 2005 American College of Radiology

n[eechtc

iclbl

iagnosis of breast lesions that are suspicious or highlyuggestive of malignancy on breast imaging is usuallychieved via either surgical excision or image-guided per-utaneous biopsy. Image-guided breast biopsy, whethererformed stereotactically or under ultrasound guidance,ffers several advantages over surgical biopsy.

Image-guided biopsy is more cost-effective than exci-ional biopsy [1-7]. Most breast lesions requiring biopsyre benign, and confirmation of a specific benign diag-

Brown Medical School, Providence, Rhode Island.

Brown University, Providence, Rhode Island.

Corresponding author and reprints: Elizabeth Lazarus, MD, Brown Medi-al School, Diagnostic Imaging, 593 Eddy Street, Providence, RI 02806;

d-mail: elazarus@lifespan.org.

88

osis by needle biopsy can obviate further surgery2,4-6,8-10]. Because surgical biopsy represents the larg-st fraction of breast cancer screening costs, replacingxcisional biopsy with image-guided biopsy can result inonsiderable cost savings. Total cost savings are evenigher when additional expenses—such as preoperativeesting, the cost of missed work, and procedural compli-ations—are considered [1,3,8,9].

For benign lesions, image-guided needle biopsy is lessnvasive and has a better cosmetic outcome and fewerhanges on follow-up breast imaging [5,8,11]. For ma-ignant lesions, surgical removal after diagnostic needleiopsy yields a higher rate of negative margins and a

ower rate of re-excision [2,4-6,7-10]. A preoperative

iagnosis of malignancy also allows for the possibility of

© 2005 American College of Radiology0091-2182/05/$30.00 ● DOI 10.1016/j.jacr.2004.10.004

nag

gfrawps

M

WdICrdab

I

OwchttAfuapgfioaituspaftdc

S

Wsb

brtpicbcRpsw

Iatggift(pmewicltt

R

OpspBbemidcbba1eiTe

Lazarus, Mainiero, Gareen/Referral for Image-Guided Breast Biopsy 489

eoadjuvant chemotherapy and for an opportunity toddress the axillary lymph nodes during a one-stage sur-ical procedure if appropriate [6,8,9,12,13].

Most patients are eligible for percutaneous image-uided biopsy, yet many still receive excisional biopsiesor diagnosis. The factors that influence a physician toecommend excisional biopsy over image-guided biopsyre often unclear. We undertook this study to identifyhether the specialty or practice type of the referringhysician is associated with a recommendation for exci-ion vs. image-guided breast biopsy.

ATERIALS AND METHODS

e retrospectively reviewed information related to theiagnostic biopsy for all women who received a Breastmaging Reporting and Data System (BI-RADS) [14]ategory 4 or 5 mammography or breast ultrasound

eport after breast imaging performed during the calen-ar year 2000 from both our academic medical centernd community-based practices. The institutional reviewoard approved this study.

maging Facilities

ur community practice is composed of five sites athich breast imaging is performed. These five imaging

enters are all located in freestanding buildings thatouse other medical practices unassociated with hospi-als. Two of these centers are located in a suburban set-ing, and the other three are located in an urban location.t these exclusively outpatient centers, radiologists per-

orm screening and diagnostic mammography and breastltrasound, as well as other diagnostic imaging tests suchs computed tomography, magnetic resonance imaging,lain film, and nuclear medicine imaging, but image-uided biopsies are not performed. The data from theseve sites were combined and compared with data fromne hospital-based breast-imaging facility that is part ofn academic medical center. At this site, percutaneousmage-guided breast biopsies are performed in additiono screening and diagnostic mammography and breastltrasound. Additionally, at the academic medical center,urgeons and radiologists participate in a multidisci-linary tumor board to discuss the diagnosis and man-gement of breast cancer cases. Patients attending theseacilities may have been referred for screening or diagnos-ic mammography, breast ultrasound, or both by aca-emic or community-based practitioners of any spe-ialty.

tatistical Analysis

e evaluated the likelihood of a patient’s undergoing aurgical excisional biopsy versus an image-guided needle

iopsy for diagnosis of a lesion seen on mammography, f

reast ultrasound, or both. We used multiple logisticegression analysis to examine the association betweenhe outcome of interest and the following covariates:hysician specialty (surgeon, obstetrician/gynecologist,nternist, or other), practice type of the referring physi-ian (academic or community based), location of thereast-imaging facility (academic or community imagingenter), patient age, insurance status of the patient, BI-ADS category, and palpability of the lesion. Data onatient race were not available retrospectively, but priortudies have shown that we serve a homogeneous, mostlyhite, population [15].After data checking and tabulation, we used SAS (SAS

nstitute, Cary, NC) to fit each multiple logistic model tossess the association between each of the covariates andhe probability of surgical biopsy as opposed to image-uided biopsy (including stereotactic, ultrasound-uided, and other unspecified core biopsies). Becausemage-guided biopsy is not a rare outcome, the odds ratiorom the logistic regression model did not approximatehe risk ratio (RR) [16]. Instead, we used GAUSSAptech Systems Inc., Maple Valley, WA) to estimate therobability of surgical excisional biopsy from the fittedodel by using the mean values in the total sample for

ach modeled covariate (for dichotomous variables, thisas the proportion of the total sample with a character-

stic). We used these fitted probabilities (risks) to directlyompute the RR estimates [17-19] and 95% confidenceimits (CL) [19]. Thus, the RR was directly estimated ashe ratio of two estimated probabilities, and this estima-ion did not rely on the rare disease assumption [16].

ESULTS

ur study population is described in Tables 1 and 2. Weerformed 39,528 mammograms and 4532 breast ultra-ound examinations at our academic and communityractices during the calendar year 2000. A total of 831I-RADS Category 4 or 5 reports (from mammography,reast ultrasound, or both) were generated from thesexaminations. Five cases (1%) were specifically recom-ended for excisional biopsy for various reasons, includ-

ng lesion location (2 cases), extensive postsurgical breastistortion limiting positioning for stereotactic biopsy (1ase), a solid mass that had been previously core needle–iopsied by using an image-guided approach that yieldedenign results which increased in size on follow-up im-ging (1 case), and a large well-circumscribed lesion in a9-year-old with a high certainty of fibroadenoma whoxpressed a preference for excision at the time of hermaging. These patients were excluded from analysis.here were 92 patients who did not have a follow-up

xcisional or image-guided biopsy and were excluded

rom analysis: 23 (3%) patients refused any biopsy or

dpol

ctuwh

490 Journal of the American College of Radiology/Vol. 2 No. 6 June 2005

eferred biopsy because of other medical issues, 33 (4%)atients had palpable lesions and underwent needle bi-psy in the referring physician’s office, and 36 (4%) wereost to follow-up.

The remaining 734 (88%) patients underwent surgi-al excisional or percutaneous image-guided biopsy (ul-rasound-guided core needle biopsy or stereotactic vac-um-assisted biopsy) for diagnosis. Of the 734 patientsho had excisional or image-guided biopsy, 723 (98%)ad data available for our final logistic regression analysis.

Table 1. Description of study population

Variable

Patients withimage-guided orexcisional biopsy

(N � 734)*Specialty of referring

physicianSurgeon 195 (25)Obstetrician/gynecologist 181 (25)Internist 292 (40)Other physician 50 (7)Not indicated 16 (2)

Location of referringphysician

Community 554 (75)Academic 164 (22)Not indicated 16 (2)

Type of biopsyExcisional 262 (36)Stereotactic 227 (30)Ultrasound 245 (33)

Age at breast-imagingstudy, y, mean (SD)

54.9 (15.0)

Insurance statusPrivate 556 (76)Medicare 106 (14)Medicaid 20 (3)Group 23 (3)Self 7 (1)

BI-RADS category4 665 (91)5 69 (9)

Palpable massYes 209 (29)No 523 (71)Not indicated 2 (1)

Location of screeningfacility

Academic 276 (38)Community 458 (62)

BI-RADS � Breast Imaging and Reporting Data System.*Data are n (%) unless otherwise noted.

The final model estimating the risk of surgical biopsy

Table 2. Detailed description of covariatesincluded in the logistic regression analysis forpatients with all information available, presentedby the location of the physician making the initialreferral for breast imaging*

Variable

Excisionalbiopsy

(n � 262)

Image-guidedbiopsy

(n � 461)

AcademicPhysician specialty

Surgeon 26 (31) 58 (69)Ob/gyn 2 (22) 7 (78)Internist 15 (23) 51 (77)Other 2 5

Physician locationCommunity 44 (33) 90 (67)Academic 39 (27) 102 (72)Not indicated 1 2

Age, y, mean (SD) 52.5 (16.5) 52.9 (12.9)Insurance status

Private 38 (31) 86 (69)Medicare 6 (25) 18 (75)Medicaid 0 (0) 9 (100)Group 1 (25) 3 (75)Self 0 (0) 3 (100)Not indicated 0 2

BI-RADS category4 44 (28) 116 (72)5 1 (17) 5 (83)

Palpable massYes 21 (45) 26 (55)No 24 (20) 95 (80)

Imaging locationCommunity 6 (25) 18 (75)Academic 39 (27) 103 (73)

CommunityPhysician specialty

Surgeon 53 (53) 46 (46)Ob/gyn 57 (37) 96 (62)Internist 52 (33) 102 (65)Not indicated 24 31Other

Physician locationCommunity 179 (41) 254 (58)Academic 6 (24) 19 (76)Not indicated 1 2

Age in years, Mean (SD) 54.1 (17.3) 56.2 (13.9)Insurance status

Private 161 (38) 265 (62)Medicare 33 (42) 45 (58)Medicaid 5 (50) 5 (50)Group 8 (58) 11 (42)Self 1 (25) 3 (75)Not indicated 9 11

BI-RADS category4 180 (36) 317 (64)5 37 (62) 23 (38)

Palpable massYes 97 (60) 66 (40)No 120 (30) 274 (70)

Imaging locationCommunity 173 (41) 250 (59)Academic 44 (33) 90 (67)

BI-RADS � Breast Imaging and Reporting Data System;Ob/gyn � obstetrics/gynecology.

*Data are n (%) unless otherwise noted.

vsisaprg

asbcC3csb9w

rCsgtdwsi

D

PaI

udtsawtb

bssmiemsgaspfc(i

bomsglbg

Lazarus, Mainiero, Gareen/Referral for Image-Guided Breast Biopsy 491

s. image-guided biopsy included four covariates: thepecialty of the physician who referred the patient for thendex breast-imaging study, the practice type of the phy-ician who referred the patient for the index breast-im-ging study, the BI-RADS category (4 or 5), and thealpability of the lesion. Table 3 presents the relativeisks comparing the probability of a surgical vs. image-uided biopsy based on the final model.

The probability of the patient’s receiving a surgical vs.percutaneous image-guided biopsy differed with the

pecialty of the physician who referred the patient forreast imaging. Patients initially referred by an obstetri-ian/gynecologist were 27% less likely (RR, 0.73; 95%L, 0.54–0.99) and women referred by an internist were5% less likely (RR, 0.65; 95% CL, 0.50–0.84) to re-eive a surgical biopsy than were women referred by aurgeon. Patients referred for their initial breast imagingy an academic physician were 37% less likely (RR, 0.63;5% CL, 0.50–0.84) to receive an excisional biopsy thanere women referred by community-based practitioners.In contrast, patients with breast-imaging studies that

eceived a BI-RADS Category 5 report were 1.44 (95%L, 1.08–1.93) times more likely to undergo an exci-

ional biopsy than were women with a BI-RADS Cate-ory 4 report. Patients with palpable lesions were almostwice as likely (RR, 1.96; 95% CL, 1.62–2.37) to un-ergo surgical biopsies than were patients whose lesionsere not palpable. The age of the patient, the insurance

tatus of the patient, and the location of the breast-imag-ng facility were not associated with the type of biopsy.

ISCUSSION

ercutaneous image-guided breast biopsy offers manydvantages over surgical excisional diagnostic biopsy.

Table 3. Results of logistic regression examining thimage-guided needle biopsy) and covariates

Variable CovariateSpecialty of referring

physician Obstetrician/gynecInternistSurgeon

Practice type of referringphysician Academic

Community-basedBI-RADS category 5

4Palpable lesion Yes

No

BI-RADS � Breast Imaging and Reporting Data System.

mage-guided needle biopsy can be performed quickly by i

sing only local anesthesia, often on the day the lesion isetected. This approach minimizes patient anxiety andime missed from work [2,4-6,8]. Performing the biopsyoon after breast imaging also decreases patient attritionnd ensures better compliance with the diagnosticorkup [4]. Image-guided biopsies are more cost-effec-

ive, and this results in cost savings in the setting of bothenign and malignant disease [1-8].

Some patients are not candidates for image-guidediopsy because of the location of their abnormality (le-ions too far posterior for a stereotactic approach or le-ions in breast tissue too thin to allow needle excursion),edical conditions that prohibit prone positioning, an

nability to cooperate with the examination, or the pres-nce of a bleeding disorder. For lesions visualized only onammography, patients with weights in excess of the

tereotactic table limit are also not eligible for image-uided needle biopsy [8,20]. Finally, lesions suggestive ofradial scar or those containing a small cluster of highly

uspicious calcifications are thought to be best ap-roached via excisional biopsy [4,8,12,21]. However,ew patients present with one of these conditions pre-luding image-guided biopsy. In our study, only 5�1%) of 889 patients were considered ineligible formage-guided biopsy.

Most of the surgeons in our area perform excisionalreast biopsy: only a few perform ultrasound-guided bi-psy, usually for palpable lesions. Therefore, surgeonsay have a direct monetary incentive to perform exci-

ional biopsy rather than to refer the patient for image-uided biopsy. This conflict of interest may explain theower likelihood of referral by surgeons for image-guidediopsy in our study. Prior research has shown that, re-arding imaging studies, physicians who self-refer for

ssociation between biopsy type (excision vs.

Risk ratio95% Confidence

limits

ogist 0.73 0.54–0.990.65 0.50–0.841.0 —

0.63 0.47–0.861.0 —1.44 1.08–1.931.0 —1.96 1.62–2.371.0 —

e a

ol

maging examinations perform more imaging examina-

tSsmcfut

finminsenrwdtetpaan

mrtfdpttssttmsb

tifweaict

mcpbatia

llorhetsm

cciatntpfipoc

ropspgpb

wsebenctopt

492 Journal of the American College of Radiology/Vol. 2 No. 6 June 2005

ions than those who refer patients to a radiologist [22].imilarly, our study suggests that physicians who canelf-refer for excisional biopsy are more likely to recom-end excisional biopsy than those who do not. In our

ommunity, very few gynecologists and no internists per-orm any type of breast biopsy, and this renders themnlikely to have any conflict of interest in referring pa-ients for biopsy.

It is also possible that some surgeons choose to per-orm excisional breast biopsy because they believe that its the better diagnostic procedure. For instance, propo-ents of surgical excision may argue that complete re-oval of the lesion removes the ambiguity that can arise

n cases (such as atypical ductal hyperplasia detected ateedle biopsy) that require excision to exclude under-ampling of ductal carcinoma in situ [23,24]. In ourxperience, some surgeons, particularly those in commu-ity practice and those trained before the advent of ste-eotactic biopsy in the 1990s, remain more comfortableith excisional biopsy than percutaneous biopsy as aiagnostic procedure. Patients referred from surgeons athe academic medical center were less likely to undergoxcisional biopsy than percutaneous biopsy as a diagnos-ic procedure. We believe that this difference is due to theresence of surgeons further specialized in breast surgeryt the academic medical center, who are more likely to beware of the benefits of percutaneous biopsy in the diag-ostic algorithm for breast disease.The association between excisional biopsy and com-unity-based practitioners of all specialties presumably

eflects the time needed for diffusion of newer medicalechnology into the community. In our practice, we per-orm all of our image-guided biopsies only at the aca-emic medical center. Therefore, community-basedractitioners have less exposure to these procedures thanheir academic counterparts. Community-based practi-ioners may also be more apt to refer their patients to aurgeon after receiving a breast-imaging report with auspicious finding, because the surgeon is perceived ashe breast specialist who will direct further diagnosis andreatment. Community-based primary care specialistsay be more likely to refer patients to community-based

urgeons, who were more likely to perform excisionaliopsy than the academic surgeons in our study.In addition to referring physician specialty and prac-

ice type, we also investigated other factors that maynfluence the referral for biopsy so that we could accountor these factors in our analysis. The fact that patientsith a Category 5 report were more likely to undergo

xcision than those with a Category 4 report may bessociated with the referring physician’s high confidencen the radiographic diagnosis and with the BI-RADSategorization system: some surgeons believe that percu-

aneous biopsy of a lesion that is highly suggestive of h

alignancy is an unnecessary diagnostic procedure be-ause the patient will require surgery anyway. However, areoperative diagnosis of malignancy by percutaneousiopsy offers the possibility of neoadjuvant chemother-py, a higher rate of negative margins after excision, andhe opportunity to address the axillary lymph nodes, andt decreases the total number of operations for appropri-te therapy [4,6-9,12,13].

In our study, patients with palpable lesions were moreikely to undergo excision than those with nonpalpableesions. This difference may be due to increased suspicionf malignancy in palpable lesions on the part of theeferring physician, or it may be related to the belief longeld by some surgeons that all palpable masses should bexcised so as not to complicate future physical examina-ions. However, image-guided biopsy has also beenhown to be valuable in the setting of palpable breastasses [6].That the location of the breast-imaging facility—

ommunity or academic medical center—was not asso-iated with biopsy type may be because all studies werenterpreted by one group of radiologists. At both ourcademic hospital and community imaging centers, weypically discuss the results of a Category 4 or 5 exami-ation with the patient and the referring physician at theime of examination. This discussion usually includesotential methods of biopsy, and image-guided biopsy isacilitated for patients whose referring physician requestst. The rate of referral for image-guided biopsy in otherractices may be different because of variations in radi-logist interactions with patients and referring physi-ians.

Long-established practice patterns on the part of theeferring physician may explain why the insurance statusf the patient did not influence biopsy choice. Also,hysicians may be unaware of their patients’ insurancetatus at the time of biopsy recommendation or mayerceive that, because both excisional biopsy and image-uided percutaneous biopsy are covered by Medicare andrivate insurance, the patient will not bear the responsi-ility for any cost difference.Our results may not be applicable to all practices,

here the patterns of referral for breast biopsy and diffu-ion of technology into the community may differ. How-ver, we have identified two possible barriers to patientseing referred for image-guided biopsy: conflict of inter-st on the part of some surgeons and slow diffusion ofewer practice patterns to community physicians. Edu-ation of both referring physicians and the public abouthe benefits of image-guided biopsy over excisional bi-psy may help to overcome these barriers. Education ofatients can be accomplished both individually at theime of the diagnostic imaging study and through public

ealth education. Radiologists can also educate referring

ptci

R

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

Lazarus, Mainiero, Gareen/Referral for Image-Guided Breast Biopsy 493

hysicians individually through discussion at the time ofhe diagnostic study or through events such as multidis-iplinary cancer management conferences and continu-ng medical education events.

EFERENCES

1. Lindfors KK, Rosenquist CJ. Needle core biopsy guided with mammog-raphy: a study of cost-effectiveness. Radiology 1994;190:217-22.

2. Liberman L, Fahs MC, Dershaw DD, et al. Impact of stereotaxic corebreast biopsy on cost of diagnosis. Radiology 1995;195:633-7.

3. Lee CH, Egglin TK, Philpotts L, Mainiero MB, Tocino I. Cost-effective-ness of stereotactic core needle biopsy: analysis by means of mammo-graphic findings. Radiology 1997;202:849-54.

4. Liberman L, Feng TL, Dershaw DD, Morris EA, Abramson AF. US-guided breast biopsy: use and cost-effectiveness. Radiology 1998;208:717-23.

5. Liberman L, Sama MP. Cost-effectiveness of stereotactic 11-gauge direc-tional vacuum-assisted breast biopsy. AJR Am J Roentgenol 2000;175:53-8.

6. Liberman L, Ernberg LA, Heerdt A, et al. Palpable breast masses: is therea role for percutaneous imaging-guided core biopsy? AJR Am J Roentge-nol 2000;175:779-87.

7. Golub RM, Bennett CL, Stinson T, et al. Cost minimization study ofimage-guided core biopsy versus surgical excisional biopsy for womenwith abnormal mammograms. J Clin Oncol 2004;22:2430-7.

8. White RR, Halperin TJ, Olson JA, Soo MS, Bentley RC, Seigler HF.Impact of core-needle breast biopsy on the surgical management of mam-mographic abnormalities. Ann Surg 2000;233:769-77.

9. Kaufman CS, Delbecq R, Jacobson L. Excising the reexcision: stereotacticcore-needle biopsy decreased need for reexcision of breast cancer. WorldJ Surg 1998;22:1023-8.

0. Smith DN, Christian R, Meyer JE. Large-core needle biopsy of nonpal-pable breast cancers. The impact on subsequent surgical excisions. ArchSurg 1997;132:256-9.

1. Huber S, Wagner M, Medl M, et al. Benign breast lesions: minimallyinvasive vacuum-assisted biopsy with 11-gauge needles—patient accep-tance and effect on follow-up imaging findings. Radiology 2003;226:783-

90.

2. Morrow M, Venta L, Stinson T, Bennett C. Prospective comparison ofstereotactic core biopsy and surgical excision as diagnostic procedures forbreast cancer patients. Ann Surg 2001;233:537-41.

3. King TA, Hayes DH, Cederbom GJ, et al. Biopsy technique has noimpact on local recurrence after breast-conserving therapy. Breast J 2001;7:19-24.

4. American College of Radiology. Breast Imaging Reporting and DataSystem (BI-RADS). 2nd ed. Reston (VA): American College of Radiol-ogy; 1995.

5. Mainiero MB, Schepps B, Clements NC, et al. Mammography-relatedanxiety: effect of preprocedural patient education. Womens Health Issues2001;11:110-5.

6. Greenland S, Thomas DC. On the need for the rare disease assumption incase-control studies. Am J Epidemiol 1982;116:547-53.

7. Wilcosky TC, Chambless LE. A comparison of direct adjustment andregression adjustment of epidemiologic measures. J Chronic Dis 1985;38:849-56.

8. Smith D. Estimating epidemiologic parameters by regression. J ChronicDis 1987;40:827-8.

9. Flanders WD, Rhodes PH. Large sample confidence intervals for regres-sion standardized risks, risk ratios, and risk differences. J Chronic Dis1987;40:697-704.

0. Jackman RJ, Marzoni FA. Stereotactic histologic biopsy with patientsprone: technical feasibility in 98% of mammographically detected lesions.AJR Am J Roentgenol 2003;180:785-94.

1. Schwartz GF, Feig SA. Nonpalpable breast lesions: biopsy methods andpatient management. Obstet Gynecol Clin North Am 2002;29:137-57.

2. Hillman BJ, Joseph CA, Mabry MR, Sunshine JH, Kennedy SD, NoetherM. Frequency and cost of diagnostic imaging in office practice—a com-parison of self-referring and radiologist-referring physicians. N EnglJ Med 1990;323:1604-8.

3. Vargas HI, Agbunag RV, Khalkhali I. State of the art of minimallyinvasive breast biopsy: principles and practice. Breast Cancer 2000;7:370-9.

4. Hall JA, Murphy DC, Hall BR, Hall KA. Open surgical biopsy fornonpalpable mammographic abnormalities: still an option compared

with core needle biopsy. Am J Obstet Gynecol 1998;178:1245-50.